From ab907f01d8e88ecded5e321018839b70d319985b Mon Sep 17 00:00:00 2001
From: Juan Cruz Viotti <jv@jviotti.com>
Date: Wed, 15 Apr 2026 20:11:57 -0400
Subject: [PATCH] Add indexing tests on schemas / collection id duplication

Signed-off-by: Juan Cruz Viotti <jv@jviotti.com>
---
 test/cli/CMakeLists.txt                       |  3 +
 .../duplicate-path-filename-matches-key.sh    | 72 +++++++++++++++++
 .../common/duplicate-path-id-matches-key.sh   | 78 +++++++++++++++++++
 ...duplicate-path-subdirectory-matches-key.sh | 77 ++++++++++++++++++
 4 files changed, 230 insertions(+)
 create mode 100755 test/cli/index/common/duplicate-path-filename-matches-key.sh
 create mode 100755 test/cli/index/common/duplicate-path-id-matches-key.sh
 create mode 100755 test/cli/index/common/duplicate-path-subdirectory-matches-key.sh

diff --git a/test/cli/CMakeLists.txt b/test/cli/CMakeLists.txt
index d646cdd61..9e9f8371f 100644
--- a/test/cli/CMakeLists.txt
+++ b/test/cli/CMakeLists.txt
@@ -98,6 +98,9 @@ if(ONE_INDEX)
   sourcemeta_one_test_cli(common index string-directory-overlap)
   sourcemeta_one_test_cli(common index non-existent-collection-directory)
   sourcemeta_one_test_cli(common index collection-path-is-file)
+  sourcemeta_one_test_cli(common index duplicate-path-filename-matches-key)
+  sourcemeta_one_test_cli(common index duplicate-path-id-matches-key)
+  sourcemeta_one_test_cli(common index duplicate-path-subdirectory-matches-key)
   sourcemeta_one_test_cli(common index rebuild-zero-to-one)
   sourcemeta_one_test_cli(common index rebuild-one-to-zero)
   sourcemeta_one_test_cli(common index resolve-schema)
diff --git a/test/cli/index/common/duplicate-path-filename-matches-key.sh b/test/cli/index/common/duplicate-path-filename-matches-key.sh
new file mode 100755
index 000000000..9e7f49816
--- /dev/null
+++ b/test/cli/index/common/duplicate-path-filename-matches-key.sh
@@ -0,0 +1,72 @@
+#!/bin/sh
+
+set -o errexit
+set -o nounset
+
+TMP="$(mktemp -d)"
+clean() { rm -rf "$TMP"; }
+trap clean EXIT
+
+mkdir -p "$TMP/schemas/foo"
+
+cat << 'EOF' > "$TMP/schemas/foo/foo.json"
+{
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "type": "object"
+}
+EOF
+
+cat << EOF > "$TMP/one.json"
+{
+  "url": "http://localhost:8000",
+  "contents": {
+    "foo": {
+      "path": "./schemas/foo"
+    }
+  }
+}
+EOF
+
+"$1" "$TMP/one.json" "$TMP/output"
+
+cd "$TMP/output"
+find . -mindepth 1 | LC_ALL=C sort > "$TMP/manifest.txt"
+cd - > /dev/null
+
+cat << 'EOF' > "$TMP/expected.txt"
+./configuration.json
+./explorer
+./explorer/%
+./explorer/%/404.metapack
+./explorer/%/directory-html.metapack
+./explorer/%/directory.metapack
+./explorer/%/search.metapack
+./explorer/foo
+./explorer/foo/%
+./explorer/foo/%/directory-html.metapack
+./explorer/foo/%/directory.metapack
+./explorer/foo/foo
+./explorer/foo/foo/%
+./explorer/foo/foo/%/schema-html.metapack
+./explorer/foo/foo/%/schema.metapack
+./routes.bin
+./schemas
+./schemas/foo
+./schemas/foo/foo
+./schemas/foo/foo/%
+./schemas/foo/foo/%/blaze-exhaustive.metapack
+./schemas/foo/foo/%/blaze-fast.metapack
+./schemas/foo/foo/%/bundle.metapack
+./schemas/foo/foo/%/dependencies.metapack
+./schemas/foo/foo/%/dependents.metapack
+./schemas/foo/foo/%/editor.metapack
+./schemas/foo/foo/%/health.metapack
+./schemas/foo/foo/%/locations.metapack
+./schemas/foo/foo/%/positions.metapack
+./schemas/foo/foo/%/schema.metapack
+./schemas/foo/foo/%/stats.metapack
+./state.bin
+./version.json
+EOF
+
+diff "$TMP/manifest.txt" "$TMP/expected.txt"
diff --git a/test/cli/index/common/duplicate-path-id-matches-key.sh b/test/cli/index/common/duplicate-path-id-matches-key.sh
new file mode 100755
index 000000000..2756fb25f
--- /dev/null
+++ b/test/cli/index/common/duplicate-path-id-matches-key.sh
@@ -0,0 +1,78 @@
+#!/bin/sh
+
+set -o errexit
+set -o nounset
+
+TMP="$(mktemp -d)"
+clean() { rm -rf "$TMP"; }
+trap clean EXIT
+
+mkdir -p "$TMP/schemas/bar"
+
+cat << 'EOF' > "$TMP/schemas/bar/test.json"
+{
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "$id": "http://localhost:8000/bar/test",
+  "type": "string"
+}
+EOF
+
+cat << EOF > "$TMP/one.json"
+{
+  "url": "http://localhost:8000",
+  "contents": {
+    "bar": {
+      "path": "./schemas/bar"
+    }
+  }
+}
+EOF
+
+"$1" "$TMP/one.json" "$TMP/output"
+
+cd "$TMP/output"
+find . -mindepth 1 | LC_ALL=C sort > "$TMP/manifest.txt"
+cd - > /dev/null
+
+cat << 'EOF' > "$TMP/expected.txt"
+./configuration.json
+./explorer
+./explorer/%
+./explorer/%/404.metapack
+./explorer/%/directory-html.metapack
+./explorer/%/directory.metapack
+./explorer/%/search.metapack
+./explorer/bar
+./explorer/bar/%
+./explorer/bar/%/directory-html.metapack
+./explorer/bar/%/directory.metapack
+./explorer/bar/bar
+./explorer/bar/bar/%
+./explorer/bar/bar/%/directory-html.metapack
+./explorer/bar/bar/%/directory.metapack
+./explorer/bar/bar/test
+./explorer/bar/bar/test/%
+./explorer/bar/bar/test/%/schema-html.metapack
+./explorer/bar/bar/test/%/schema.metapack
+./routes.bin
+./schemas
+./schemas/bar
+./schemas/bar/bar
+./schemas/bar/bar/test
+./schemas/bar/bar/test/%
+./schemas/bar/bar/test/%/blaze-exhaustive.metapack
+./schemas/bar/bar/test/%/blaze-fast.metapack
+./schemas/bar/bar/test/%/bundle.metapack
+./schemas/bar/bar/test/%/dependencies.metapack
+./schemas/bar/bar/test/%/dependents.metapack
+./schemas/bar/bar/test/%/editor.metapack
+./schemas/bar/bar/test/%/health.metapack
+./schemas/bar/bar/test/%/locations.metapack
+./schemas/bar/bar/test/%/positions.metapack
+./schemas/bar/bar/test/%/schema.metapack
+./schemas/bar/bar/test/%/stats.metapack
+./state.bin
+./version.json
+EOF
+
+diff "$TMP/manifest.txt" "$TMP/expected.txt"
diff --git a/test/cli/index/common/duplicate-path-subdirectory-matches-key.sh b/test/cli/index/common/duplicate-path-subdirectory-matches-key.sh
new file mode 100755
index 000000000..8c5a9f069
--- /dev/null
+++ b/test/cli/index/common/duplicate-path-subdirectory-matches-key.sh
@@ -0,0 +1,77 @@
+#!/bin/sh
+
+set -o errexit
+set -o nounset
+
+TMP="$(mktemp -d)"
+clean() { rm -rf "$TMP"; }
+trap clean EXIT
+
+mkdir -p "$TMP/schemas/baz/baz"
+
+cat << 'EOF' > "$TMP/schemas/baz/baz/test.json"
+{
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "type": "array"
+}
+EOF
+
+cat << EOF > "$TMP/one.json"
+{
+  "url": "http://localhost:8000",
+  "contents": {
+    "baz": {
+      "path": "./schemas/baz"
+    }
+  }
+}
+EOF
+
+"$1" "$TMP/one.json" "$TMP/output"
+
+cd "$TMP/output"
+find . -mindepth 1 | LC_ALL=C sort > "$TMP/manifest.txt"
+cd - > /dev/null
+
+cat << 'EOF' > "$TMP/expected.txt"
+./configuration.json
+./explorer
+./explorer/%
+./explorer/%/404.metapack
+./explorer/%/directory-html.metapack
+./explorer/%/directory.metapack
+./explorer/%/search.metapack
+./explorer/baz
+./explorer/baz/%
+./explorer/baz/%/directory-html.metapack
+./explorer/baz/%/directory.metapack
+./explorer/baz/baz
+./explorer/baz/baz/%
+./explorer/baz/baz/%/directory-html.metapack
+./explorer/baz/baz/%/directory.metapack
+./explorer/baz/baz/test
+./explorer/baz/baz/test/%
+./explorer/baz/baz/test/%/schema-html.metapack
+./explorer/baz/baz/test/%/schema.metapack
+./routes.bin
+./schemas
+./schemas/baz
+./schemas/baz/baz
+./schemas/baz/baz/test
+./schemas/baz/baz/test/%
+./schemas/baz/baz/test/%/blaze-exhaustive.metapack
+./schemas/baz/baz/test/%/blaze-fast.metapack
+./schemas/baz/baz/test/%/bundle.metapack
+./schemas/baz/baz/test/%/dependencies.metapack
+./schemas/baz/baz/test/%/dependents.metapack
+./schemas/baz/baz/test/%/editor.metapack
+./schemas/baz/baz/test/%/health.metapack
+./schemas/baz/baz/test/%/locations.metapack
+./schemas/baz/baz/test/%/positions.metapack
+./schemas/baz/baz/test/%/schema.metapack
+./schemas/baz/baz/test/%/stats.metapack
+./state.bin
+./version.json
+EOF
+
+diff "$TMP/manifest.txt" "$TMP/expected.txt"